[SCIP] Most Infeasible Branching for MIQCP

[Timo Berthold]

Dear Ahmed,

> So does the most
> infeasible branching in SCIP consider only integrality constraints
Yes.
> MIQCP or does it consider quadratic constraints too?
No.

Per default, branching on violated nonconvexities will only happen when
all integer variables take integral values in the current LP solution.
Thus, as long as most infeasible (or other LP-based  branching rules like
relpscost) finds a violated integer to branch on, it will do so. If,
however, the LP happens to be integral, then most infeasible will branch
on the most violated nonconvexity.

Hope that helps.

Best regards,
Timo


> Hi All,
>
> According to my my understanding, most infeasible branching is an LP
> branching technique and selects the integer variable whose fractional
> component is closest to 0.5. Is it the same for an MIQCP? The reason I'm
> asking this is that in MIQCP linear outer approximation is used for
> quadratic constraints. Hence some quadratic constraints could also be
> violated besides integrality constraints can be violated. So does the most
> infeasible branching in SCIP consider only integrality constraints in
> MIQCP or does it consider quadratic constraints too?
>
> Regards,
> Ahmed
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip
>